.jpg)
10 March 2025
Artificial intelligence and automation are at the heart of current digital transformations. Supported by innovative solutions like our Sparkneo solution, they offer fascinating perspectives while raising essential questions: how are these technologies revolutionizing content creation? Are they in a position to meet ever more complex needs? And above all, what opportunities are available to businesses that adopt them?
During the Paris Retail Week 2024 workshop on automation and personalized AI, led by Iban Touchet, director of Incremys & Sparkneo, and Yoan Lavrentieff, sales manager, you will discover how these technologies can transform the way businesses generate content, responding to growing market demands.
1. The State of AI So Far
a. Generic AI vs Custom AI
SEO is entering a new era with the advent of AI, which has considerable impacts on content creation. Basically, this technology is designed to generate texts in an automated way. However, it is crucial to make a clear distinction between two types of AI: generic AI and personalized AI.
THEGeneric AI, which many are familiar with, is often characterized by a lack of brand identity. The training data used to generate content is not always controlled by the user, which means that the texts produced may lack relevance and specificity. Moreover, this approach does not usually follow a particular brief or specific requests, resulting in the creation of very generic content that then requires manual rewriting. This is typically what you experience when using tools such as Playground ChatGPT to generate text, which you then have to rework by hand.
On the other hand, custom AI is designed to produce directly publishable content. When you use this technology within Incremys, the engines are built according to 3 main principles:
- Your brand identity: the texts generated are in perfect harmony with the image you want to project.
- Your data: the engine considers “hot” data sources such as URLs, files, etc... thus ensuring the relevance of the generated texts.
- Your brief: all instructions, from topics to be addressed to the desired network, are taken into account.
So, if you give instructions for a given keyword, the content generated by custom AI will respect all the guidelines provided.
For our part, when a new Incremys customer is signed, meticulous work is done by the marketing department to position the engine in such a way that it is in line with the brand.
This approach guarantees unique and publishable content, very different from what we have been able to achieve with generic generative AIs up until now. The quality and relevance of content produced with custom AI is much higher because it is designed to meet specific and strategic needs.
b. Custom AI illustration
We have integrated this innovative technology into several essential functionalities of our solution. Incremys. To illustrate the concept of personalized AI, we offer you a concrete example: Building a brainstorm (overview generated from a keyword to make it easier for managers to write content), followed by the creating a table of contents, then the generating content for a category page. This practical case was discussed in our article on SEO and content creation through artificial intelligence. For more information, we invite you to consult the chapter”Personalized AI: what are the uses for SEO?”.
2. Large-scale automation
a. Originally: AI as an assistant
Historically, Incremys' AI was designed as a helper in a text editor several years ago. The editor allows users to generate texts intuitively and efficiently. However, over time, it became clear that, although the quality of the texts generated was satisfactory, there was a major problem: the transition to the ladder.
For a business that needs to create a few hundred or thousands of local pages, or even thousands of product pages, using one person to click a button hundreds of times is not economically viable. This observation led to a thorough reflection on the need for a solution. of automation.
A request from our customers to cope with a large volume
Our automation module was co-built with our customers, who expressed specific needs in terms of content creation. These users were already happy with our ability to generate single texts, but the aim was to go far beyond that. The idea was not only to produce a text, but to generate 10, 100, or even 1000 texts with a single click. Automation responds to a real market demand, allowing businesses to deal with growing volumes of content without compromising quality.
b. How does it work?
Understanding how automation works is based on several key points that respond directly to customer demands.
Creation of an AI brief: First of all, the first step is to create briefs adapted to different types of content. For example, users can create a brief for a product page, another for a Category page, one for one Local page, and one for the Distribution Marketplace. Each brief allows you to define the template and the type of text you want to generate. This step is critical, as a well-formulated brief will guide content creation in an effective manner.
Definition of a data strategy: custom AI is designed to take into account hot data, thus minimizing the risk of generation models hallucinating. When creating text in an editor, it is usually possible to review and correct the content. However, when one aims to generate 10,000 texts, it becomes impractical to reread each element. That's why a solid data strategy is essential, ensuring that the texts produced are of high quality and relevant.
Creation of projects from hot data: the projects are done on the spot by simply sending the data expected by a brief. This operation makes it possible to launch projects according to arrivals. For example, if in January, a new start of 1,000 products is expected, the user can simply launch a project to generate 1,000 texts corresponding with a brief of the product pages. In the same way, if a new series of products arrives the following month, another project can be initiated to produce the necessary texts always with the same brief of the product pages.
Thus, by combining the creation of structured briefs, rigorous data control and fluid project management, automation becomes a powerful tool in the service of content creation on a large scale.
c. Focus on data strategies
Our initial approach is to allow users to easily import data via CSV files that is to say shopping flows. Overall, these are attribute columns and each row corresponds to a product or a category. We have also integrated API connections to facilitate the direct import of catalogs into our platform, allowing users to generate content quickly and effortlessly.
The problem is that some customers do not have their data in these formats.
So we allowed the hot scraping, which allows attributes to be extracted directly from URL, whether internal or external. For example, when a customer receives a pack of 50 mowers, they can use the Manufacturer URLs to retrieve the attributes necessary for text generation, even if it does not have this data internally.
This method does not consist in copying texts or graphics, but in extract fundamental data web pages. This ensures information integrity while allowing flexibility in content creation.
Finally, our latest evolution consists ofextracting from files. This case is useful when, for example, a customer wants to generate product pages from PDF, PPTX, DOC documents supplied by the manufacturers. We have also integrated technology Vision AI, allowing attributes to be extracted directly from Of images. For example, this technology can identify and describe a blue dress from a photo.
These data strategies are designed to offer maximum flexibility and allow businesses to generate content quickly. relevant and of high quality, adapted to their specific needs.
D. Typical use cases
Use cases start with our core business: SEO optimization via the Incremys solution. Integrated as an automation module, it makes it possible to generate at scale product pages, of category pages, of local pages and content for the Marketplace distribution.
We have expanded our scope of application with the possibility of creating content such as CRM. In this context, AI can generate hyper-personalized emails, based on the attributes of your customer relationship management system.
Another application case is the creation of content dedicated to media. Currently, we are in a position to generate content for various advertisements, in particular for facebook, Google Ads, and formats of Display Or of shopping. This represents another type of generated content, always in line with customer needs.
It is important to emphasize that the fundamental remains the same: with a personalized AI that masters your Brand tone, your data and who follows your instructions, it becomes possible to generate adapted texts, whether it is a Web page, of a mail, or of a publicity. The only difference is in the Brief, which determines the specific nature of the content to be produced.
Our Sparkneo and Incremys users can, by creating adapted briefs and controlling their data, produce tens, hundreds, even thousands of publishable contents in a few clicks!
To further explore how theAutomation can transform your content strategy, do not hesitate to Make contact with our team. A demo session will allow you to discover the specific functionalities of our solution and to identify how it can meet your specific needs:
3. To go further - Projects on demand
The use cases we presented are mainly standard, outside of the platform, we also carry out other types of projects. These initiatives go beyond current applications and allow for the exploration of solutions tailored to unique needs. What we have just dealt with is only the The tip of the iceberg, hinting at the multitude of opportunities that personalized AI can offer your business.
a. Takeover and improvement of all the existing blog
The improvement The entirety of a blog or magazine can be carried out completely automatically as part of a project. Although it is possible to improve a text one by one via the text editor Incremys, our approach Sparkneo allows you to process the entire blog/magazine.
The steps in this process are as follows:
- Massive import : first, all of your pages and content are integrated en masse into the platform.
- Keyword detection : custom AI is able to detect the keyword on each page in an intelligent way. Unlike some solutions that claim to provide a target keyword, our system guarantees accurate identification of keywords per page.
- Improvement decision : then, you have to agree on the possibility of triggering the improvement or not. For example, if a page ranks first on Google, it's best not to change it, it's a rule of thumb.
- Automated improvement : If the improvement is triggered, the custom AI can perform multiple actions. It can optimize tags, add content, and provide value by drawing on external data sources.
- Integration : due to the scale of operations, it is crucial that texts are generated in an integration format. This makes it possible to publish and update automatically in bulk, with the possibility of managing 1000, 2000 or even 5000 blog or magazine articles.
b. Generate blog posts or FAQ pages at scale
The second type of project concerns large-scale generation of blog posts and FAQ pages. In this context, it is mandatory to standardize processes and to have a structured brief that applies to all generated content.
As soon as a Permutation around a keyword is possible, it is relevant to consider automation to quickly produce personalized content based on available data.
For the generation of blog posts, for example, if we focus on the garden center sector, it is possible to base content on keyword permutations. Topics such as “what to plant in January?” ”, “what to plant in February?” ”, and so on, are examples of keywords that lend themselves to this approach.
When it comes to FAQ pages, a typical example is that of a customer with 10,000 brands in a huge marketplace. For each brand, a page is created, including information about the brand and its flagship products.
The steps in this process are as follows:
- Define a content strategy : a permutation that is triggered around a common structure and themes.
- Define brief templates : definition of the brief templates for each keyword for the type of content you want to obtain.
- Automatic generation : content generation that is launched according to your tone of voice, your brief and your data.
- Integration : due to the scale of operations, it is crucial that texts are generated in an integration format. This makes it possible to automatically publish and update in bulk.
c. Other type of tailor-made project: non-standard projects
Volumetry projects: in the context of high-volume projects, it is common to process requests involving 20, 30, or even 40 millions of words. Generally, these projects take place outside the platform. For example, it may be the generation of 6000 pages in 15 languages, with a few hundred words per page. Customers involved in this type of project often adopt a turnkey service mode.
Projects at Data Complexe: another type of project concerns situations of extreme data complexity. For example, when working with franchises, each franchisee may have their own website, managed by their own agency, containing specific and personalized information. In this case, when a central entity wishes to create a global site bringing together the information of all its franchisees, it faces a major difficulty:These data sources come from websites with varied structures, making it difficult to accurately identify major information.
Some may have 200 pages, others 30, 20, or even 5 pages. In these situations, data strategy becomes particularly complex, especially when information is not available internally or easily accessible externally. This requires a project approach that is more personalized and adapted to the specific needs of the client.
Media projects: There is a variety of content to be generated when it comes to media projects. For example, for a specific commercial offer, such as a promotion offering 30% off the third item purchased, it may be necessary to create 15, 20, or even 30 different texts adapted to various communication channels.
In this context, the main difficulty lies in the respect for the character constraints imposed by each advertising channel. This is an interesting challenge, because you have to be able to produce texts that respect these constraints while meeting market expectations.
Conclusion
This workshop demonstrated how custom AI and automation can transform content creation. Personalized AI is a powerful tool that not only generates quality content, but also optimizes business time and resources.
By adopting these technologies, it becomes possible to produce relevant content on a large scale, responding to growing market challenges.
Concrete example
.svg)
.svg)
.jpg)
.svg)
.avif)